From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-wm0-f71.google.com (mail-wm0-f71.google.com [74.125.82.71]) by kanga.kvack.org (Postfix) with ESMTP id 2CB236B0009 for ; Mon, 26 Mar 2018 10:36:36 -0400 (EDT) Received: by mail-wm0-f71.google.com with SMTP id v131so4099405wmv.6 for ; Mon, 26 Mar 2018 07:36:36 -0700 (PDT) Received: from metis.ext.pengutronix.de (metis.ext.pengutronix.de. [2001:67c:670:201:290:27ff:fe1d:cc33]) by mx.google.com with ESMTPS id r28si544989wra.382.2018.03.26.07.36.34 for (version=TLS1_2 cipher=ECDHE-RSA-CHACHA20-POLY1305 bits=256/256); Mon, 26 Mar 2018 07:36:34 -0700 (PDT) Message-ID: <1522074988.1196.1.camel@pengutronix.de> Subject: Re: [RFC] Per file OOM badness From: Lucas Stach Date: Mon, 26 Mar 2018 16:36:28 +0200 In-Reply-To: <20180130102855.GY21609@dhcp22.suse.cz> References: <20180118170006.GG6584@dhcp22.suse.cz> <20180123152659.GA21817@castle.DHCP.thefacebook.com> <20180123153631.GR1526@dhcp22.suse.cz> <20180124092847.GI1526@dhcp22.suse.cz> <583f328e-ff46-c6a4-8548-064259995766@daenzer.net> <20180124110141.GA28465@dhcp22.suse.cz> <36b49523-792d-45f9-8617-32b6d9d77418@daenzer.net> <20180124115059.GC28465@dhcp22.suse.cz> <60e18da8-4d6e-dec9-7aef-ff003605d513@daenzer.net> <20180130102855.GY21609@dhcp22.suse.cz> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: owner-linux-mm@kvack.org List-ID: To: Michal Hocko , Michel =?ISO-8859-1?Q?D=E4nzer?= Cc: linux-kernel@vger.kernel.org, amd-gfx@lists.freedesktop.org, Christian.Koenig@amd.com, linux-mm@kvack.org, dri-devel@lists.freedesktop.org, Roman Gushchin Hi all, Am Dienstag, den 30.01.2018, 11:28 +0100 schrieb Michal Hocko: > On Tue 30-01-18 10:29:10, Michel DA?nzer wrote: > > On 2018-01-24 12:50 PM, Michal Hocko wrote: > > > On Wed 24-01-18 12:23:10, Michel DA?nzer wrote: > > > > On 2018-01-24 12:01 PM, Michal Hocko wrote: > > > > > On Wed 24-01-18 11:27:15, Michel DA?nzer wrote: > > > > > > [...] > > > > > > 2. If the OOM killer kills a process which is sharing BOs > > > > > > with another > > > > > > process, this should result in the other process dropping > > > > > > its references > > > > > > to the BOs as well, at which point the memory is released. > > > > > > > > > > OK. How exactly are those BOs mapped to the userspace? > > > > > > > > I'm not sure what you're asking. Userspace mostly uses a GEM > > > > handle to > > > > refer to a BO. There can also be userspace CPU mappings of the > > > > BO's > > > > memory, but userspace doesn't need CPU mappings for all BOs and > > > > only > > > > creates them as needed. > > > > > > OK, I guess you have to bear with me some more. This whole stack > > > is a > > > complete uknonwn. I am mostly after finding a boundary where you > > > can > > > charge the allocated memory to the process so that the oom killer > > > can > > > consider it. Is there anything like that? Except for the proposed > > > file > > > handle hack? > > > > How about the other way around: what APIs can we use to charge / > > "uncharge" memory to a process? If we have those, we can experiment > > with > > different places to call them. > > add_mm_counter() and I would add a new counter e.g. MM_KERNEL_PAGES. So is anyone still working on this? This is hurting us bad enough that I don't want to keep this topic rotting for another year. If no one is currently working on this I would volunteer to give the simple "just account private, non-shared buffers in process RSS" a spin. Regards, Lucas